Fugitive Dust Emissions from Off-Road Vehicle Maneuvers on Military Training Lands

FUGITIVE DUST EMISSIONS FROM OFF-ROAD VEHICLE MANEUVERS ON MILITARY TRAINING LANDS

J. C. Meeks, L. E. Wagner, R. G. Maghirang, J. Tatarko, N. Bloedow

ABSTRACT. Off-road vehicle training can contribute to air quality degradation because of increased wind erosion as a result of soil disruption during high wind events. However, limited information exists regarding the impacts of off-road vehicle maneuvering on wind erosion potential of soils. This study was conducted to determine the effects of soil texture and intensity of training with off-road vehicles on fugitive dust emission potential due to wind erosion at military training installations. Multi-pass military vehicle trafficking experiments involving wheeled and tracked vehicles were conducted at three military training facilities (Fort Riley, Kansas; Fort Benning, Georgia; and Yakima Training Center, Washing- ton) with different vegetative conditions and soil textures. The top 6 cm of soil was collected with minimum disturbance into trays and tested in a laboratory wind tunnel for dust emission potential. In wind tunnel testing, the amount of emitted dust was measured using a Grimm aerosol spectrometer. The dust emission potential due to wind erosion was significantly influenced by soil texture, vehicle type, and number of passes. For the light wheeled vehicle, total dust emissions (<20 μm) increased by 357% and 868% for 10 and 50 passes, respectively, from the undisturbed soil condition. For the tracked vehicle, an average increase in total dust emissions (<20μm) of 569% was observed between undisturbed soil and one pass, with no significant increase in emission potential beyond one pass. For the heavy wheeled vehicle, evaluated only at Yakima, emissions (<20 μm) increased by 2,108% and 5,276% for 10 and 20 passes, respectively, from the undisturbed soil condition. Soil texture also played an important role in dust emission potential. For all treatment effects with the light wheeled vehicle, there was a 1,396% increase in emissions (<20 μm) on loamy sand soil over silty clay loam soil. Keywords. Air quality, Particulate matter, Soil, Wind erosion.

he U.S. Department of Defense (DoD) manages of vegetation, soil compaction, and soil loss due to erosion and conducts training on over 12 million ha of by water and wind (Althoff et al., 2010). Although some land (DoD, 2013). This land is used for various windborne dust events can be classified as exceptional purposes, including residential and commercial events, soil transport by wind movement can contribute to Tactivities and intensive combat training operations. Many monitored particulate matter (PM) values that exceed the combat training activities require navigating large, heavy National Ambient Air Quality Standards (NAAQS) (Ash- off-road vehicles across potentially sensitive and undis- baugh et al., 2003). Suspended wind-eroded particulate turbed off-road locations. These activities can create signif- matter (PM) emitted from military installations can be car- icant disturbance to the soil surface and adversely affect the ried long distances from the training land and well beyond local ecosystem. The most prevalent impacts include loss property boundaries. Soil erosion by wind is a dynamic process that often results in loss of the fertile top layer of soil, which can reduce Submitted for review in September 2013 as manuscript number NRES agricultural production as well as degrade local air quality 10428; approved for publication by the Natural Resources & (Skidmore and van Donk, 2003; Diaz-Nigenda et al., 2010). Environmental Systems of ASABE in December 2014. Presented at the Intensive tillage and other agricultural practices on arid and 2013 ASABE Annual Meeting as Paper No. 131508778. Contribution No. 13-352-J from the Kansas Agricultural Experiment semi-arid land can contribute to an increase in soil loss due Station. to wind erosion. Many other factors, such as unpaved Mention of company or trade names is for description only and does roads, confined animal operations, and intensive use as not imply endorsement by the USDA. The USDA is an equal opportunity military training lands, may also contribute to dust emis- provider and employer. The authors are Jeremy C. Meeks, Former Graduate Student, sions due to wind erosion (Gillies et al., 2005). Department of Biological and Agricultural Engineering, Kansas State In addition to simulation and field studies, previous re- University, Manhattan, Kansas; Larry E. Wagner, ASABE Member, search on wind erosion has also used wind tunnels under Agricultural Engineer, USDA-ARS Agricultural Systems Research Unit, field and laboratory conditions. Several studies (Van Pelt et Fort Collins, Colorado; Ronaldo G. Maghirang, ASABE Member, Professor, Department of Biological and Agricultural Engineering, Kansas al., 2010; Sweeney et al., 2008; Marticorena and Bergamet- State University, Manhattan, Kansas; John Tatarko, Soil Scientist, ti, 1997) have used in situ semi-portable wind tunnels that USDA-ARS Agricultural Systems Research Unit, Fort Collins, Colorado; are set up directly in the field on the surface of interest. Nicholas Bloedow, Graduate Research Assistant, Department of Statistics, Kansas State University, Manhattan, Kansas. Corresponding author: Other studies have used large laboratory wind tunnels to Larry Wagner, USDA-ARS Agricultural Systems Research Unit, 2150 control both wind speed and soil surface conditions to some Centre Ave., Bldg. D, Suite 200, Fort Collins, CO 80526; phone: 970-492- degree (Guoliang et al., 2003; Kohake et al., 2010). These 7382; e-mail: [email protected].

Transactions of the ASABE Vol. 58(1): 49-60 2015 American Society of Agricultural and Biological Engineers ISSN 2151-0032 DOI 10.13031/trans.58.10428 49 studies have shown that wind tunnels can be used success- fully to study the underlying principles of wind erosion while simulating the optimal conditions that contribute to initiating large wind erosion events. Limited information exists regarding the impacts of military training activities on the wind erosion potential of soil. This research was conducted to determine the effects of off-road trafficking on fugitive dust emissions from military training lands. Specifically, the overall objective was to determine the effects of soil texture, vehicle type, and number of passes on dust emission potential using a laboratory wind tunnel.

MATERIALS AND METHODS FIELD SITES Testing was conducted at three DoD military training sites: Fort Riley, Kansas; Fort Benning, Georgia; and Ya- kima Training Center, Washington. These three sites represent a diversity of vegetative conditions, soil textures, and off-road training activities. Two off-road military training vehicles were used at each site, representing a range of relatively lightweight to very heavy fully armored military off-road vehicles. Military training facilities cover a wide variety of locations within the U.S. Each experimental site was selected to provide a variety of soil textures as well as represent wind erosion susceptible soils at military installations in different climates, but not necessarily represent a majority of the soils at all U.S. military training facilities. Fort Riley (39° 18′ 0″ N; 96° 55′ 18″ W) was the first installation chosen for testing, which was completed in Figure 1. (top) Heavy tracked model M1A1 and (bottom) light ar- October and November 2010. This installation is located in mored HMMWV model M1025A2 vehicles. the Flint Hills region of north central Kansas in Geary and Riley counties and covers 407 km2, with 287 km2 used for and on 8-12 November 2010 with the HMMWV. maneuver training (EPA, 2010). Fort Riley receives annual The second installation was Fort Benning, Georgia (32° precipitation of 889 mm and has an annual mean tempera- 24′ 14″ N; 84° 45′ 21″ W), with testing conducted on 21-24 ture of 12.6°C. Testing at Fort Riley was conducted on silty July 2012. Fort Benning is located on 765 km2 in west cen- clay loam (SiCL) and silt loam (SiL) soil textures (table 1). tral Georgia. This site is located in the humid Southeastern An M1A1 Abrams heavily armored tracked vehicle region of the U.S., which has relatively high precipitation (61,500 kg) and a light armored M1025A2 high-mobility and is affected by tropical maritime events, including fre- multipurpose wheeled vehicle (3,100 kg), commonly called quent thunderstorms and inland-moving hurricanes a Humvee (HMMWV), were used for testing. Figure 1 (NCDC, 2011). Fort Benning has average annual precipita- shows the two vehicles during testing at the Fort Riley site. tion of 1245 mm and a mean annual temperature of 18.3°C. The vegetation at the Fort Riley site was a mixed tallgrass Testing at Fort Benning was done at Rowan Hill on a prairie dominated by big bluestem (Andropogin gerardii) loamy sand (LS) soil texture (table 1). The two vehicles and Indian grass (Sorghastrum nutans) species. Testing on used were a HMMWV up-armored vehicle (model both soils was done on 18-22 October 2010 with the M1A1 M1151A1) with a curb weight of approximately 3,800 kg

Table 1. Measured soil properties for the original surface at the experimental sites. Sand Silt Clay Organic Matter IWC[a] IBD[a] Site Soil Texture (g kg-1) (g kg-1) (g kg-1) (g kg-1) (kg kg-1) (g cm-3) Silt loam 0.16 T[b] 102 707 190 40 1.27 (SiL) 0.17 W Fort Riley Silty clay loam 0.18 T 79 641 280 41 1.22 (SiCL) 0.21 W Fort Benning Loamy sand 877 95 28 9 0.02 1.47 (Rowan Hill) (LS) Loam Yakima Training Center 513 438 50 21 0.01 1.30 (L) [a] IWC = initial water content and IBD = initial bulk density (5 cm depth) taken prior to trafficking at each site. [b] T denotes the tracked vehicle experiment (Oct. 2010) and W denotes the wheeled vehicle experiment (Nov. 2010) at Fort Riley.

50 TRANSACTIONS OF THE ASABE Table 2. Vehicle specifications and locations used in experiments on specified dates. Traction Weight Track/Tire Width Speed Vehicle Type (kg) (cm) (km h-1) Fort Riley Fort Benning Yakima M1025A2 Wheeled 3,100 4.72 25-30 Nov. 2010 - Aug. 2012 HMMWV M1A1 Tracked 61,500 9.45 8 Oct. 2010 July 2012 - Abrams tank M1151A1 Wheeled 3,800 4.72 25-30 - July 2012 - Up-armored HMMWV M925A1 Wheeled 15,000 5.90 12 - - Aug. 2012 Five-ton fire truck and an M1A1 heavily armored tracked vehicle weighing heavier vehicles. Additionally, the M1A1 Abrams tank is approximately 61,500 kg. Fort Benning is within the long- referred to as “tracked vehicle,” whereas the M925A1 wa- leaf pine ecosystem, and the site was dominated by loblolly ter tanker fire truck is referred to as “heavy wheeled vehi- pine (Pinus taeda), longleaf pine (Pinus palustris), and cle.” Table 2 summarizes the vehicle properties. white oak (Quercus alba), with areas of herbaceous To accommodate the type of turns that military vehicles groundcover including grasses dominated by thin paspalum normally make during training maneuvers, a figure-8 pat- (Paspalum setacum) and southern crabgrass (Digitaria cil- tern (fig. 2) on 80 m × 40 m plots was followed, similar to iaris) and forbs dominated by orangegrass (Hypericum gen- those employed by Althoff and Thien (2005). The inside tianoides) and pine barren tricklefoil (Desmonium turning radius of the vehicles within the figure-8 plots was strictum). approximately 10 m. In most training maneuvers, a point Yakima Training Center, Washington (46° 41′ 36″ N; on the ground may experience many passes, or high traffic 120° 26′ 15″ W), was the third site, with testing conducted intensity, by one or more vehicles. High traffic intensity on 11-14 August 2012. This site is situated on 1323 km2 of also leads to increased damage and the potential for more shrub-steppe land in south central Washington. This site is dust emissions. located in the Yakima Valley region with a mean annual The number of trafficking passes was established for temperature of 12.7°C and is dry, with average annual pre- each vehicle based on test runs representing minimum, me- cipitation of 207 mm (NCDC, 2011). Testing at Yakima dium, and severe relative levels of disturbance (i.e., traf- Training Center was conducted on a loam (L) soil texture ficking intensity levels) that would also likely provide (table 1). A tracked vehicle was not available for use; there- measurable differences among the trafficking intensity lev- fore, a fully loaded, tandem dual rear axle, six-wheel drive els (Retta et al., 2013). The number of vehicle passes was M925A1 water tanker fire truck weighing 15,000 kg was kept the same for each vehicle type at all experimental used in addition to a light armored HMMWV (model sites. For the M1A1, the three trafficking intensity levels M1025A2). Vegetation at the Yakima Training Center site consisted of one pass, followed by four additional passes was characterized by a shrub-steppe vegetation dominated (five cumulative passes), and then followed by five final by perennial bunchgrasses such as bluebunch wheatgrass passes (10 cumulative passes). For the HMMWV, the traf- (Pseudoroegneria spicata) and Sandberg bluegrass (Poa ficking intensity levels consisted of 10, 25, and 50 cumula- secunda). tive passes. The M925A1 trafficking levels consisted of Initial (untrafficked) soil samples were taken to a 5 cm double the M1A1 passes, for cumulative totals of 2, 10, and depth. Bulk density and water content (Retta et al., 2013) 20 passes. The M1A1 was driven at a typical off-road ma- were measured to provide a general characterization of the neuvering speed of 8 km h-1, the HMMWV at approximate- soils and are reported in table 1. Clay contents of the ly 25 to 30 km h-1, and the M925A1 at approximately <2 mm soil were measured by pipette method, sand frac- 12 km h-1. However, the speed was adjusted at the discre- tion by wet sieving, and silt by difference according to the tion of the driver, with a primary consideration of staying method of Gee and Dani (2002). Organic matter was meas- within the same vehicle tracks for every pass. ured at the Kansas State University Soil Testing Laboratory The figure-8 patterns were staked out for each treatment using the Modified Walkley-Black method (Swift, 1996). and replication at each installation (fig. 2). Three replica- Percent vegetation cover was determined by the line inter- tions were conducted for each vehicle/soil treatment com- cept method (Sloneker and Moldenhauer, 1977). bination at each site. Three levels of trafficking intensity At each of the installations, sites were selected with as little prior disturbance as possible. The sites may have been used for training activities in the past, but all locations had been undisturbed by vehicles for an extended period of time, e.g., several years, with plentiful natural vegetation regrowth for the site’s climatic conditions. At the Fort Ri- ley site, the vegetation was so dense that it required mow- ing and removal prior to sampling. Note that the two models of HMMWV are referred to as “light wheeled vehicle” and were considered the same for analysis, as the differences in weight were minor compared with the other much Figure 2. Figure-8 plot showing tray sample locations: SS = straight section, CI = curve inside, and CO = curve outside.

58(1): 49-60 51 (i.e., number of passes) were conducted on each replicated figure-8, with samples taken after the completion of each trafficking intensity level as well as before any trafficking to determine initial conditions. Wind tunnel tray samples were collected on the outside and inside curved sections (CO and CI) as well as from a straight section (SS). Figure 2 illustrates those sample locations within the figure-8 plots. All sampling locations per treatment are listed in the sampling matrix in table 3, which indicates the location within the figure-8 (i.e., CO, CI, and SS) at which samples were collected and the number of passes for each vehicle at each site. Due to resource con- straints, a balanced experiment was not conducted on the Fort Riley site with the HMMWV.

WIND TUNNEL TESTING The top 6 cm of the soil surface was carefully removed from a 122 cm × 20 cm area for each sample location. Samples were removed with a flat-bottom shovel and placed into trays (fig. 3) to minimize disturbance. These trays allowed consistent samples to be collected and also aided in ease of transportation, storage, and testing. The trays were retrieved from the field and stored on closely spaced shelving in a trailer for transport from the field site to the laboratory in Manhattan, Kansas, to minimize disturbance during transportation. It was noticed that some Figure 4. Slot sampler: (left) side view and (right) front view. settling and shifting occurred during transport of some soil trays; however, the disturbance was minimal and did not close to 100% sampling efficiency for all particle sizes for significantly alter the tray configuration. All samples were the test conditions. The isokinetic conditions ensure that the allowed to completely air-dry in a greenhouse for at least sampler does not interfere with the movement of the air one month prior to wind tunnel testing. stream and allow for more accurate measurement of partic- A vertically integrated, slot-style sampler (fig. 4), devel- ulate emissions. The sampler has a 5 mm wide slot for par- oped by Mirzamostafa et al. (1998), was used for wind tun- ticles to enter the system. The system also contains a cy- nel testing. The primary advantage of this sampler is that it clonic design inside, so saltation size or larger particles can operate isokinetically at various wind speeds, providing

Table 3. Sampling matrix for all triple replicated treatments. M1A1 (Tracked Vehicle) M925A1 (Heavy Wheeled Vehicle) Fort Benning Fort Riley Yakima Training Center Sample Number of Passes Sample Number of Passes Sample Number of Passes Location 0 1 5 10 Location 0 1 5 10 Location 0 2 10 20 CO[a] X X X X CO X[b] X X X CO X X X X CI X X X X CI X[b] X X X CI X X X X SS X X X X SS X[b] X X X SS X X X X HMMWV (Light Wheeled Vehicle) Fort Benning Fort Riley Yakima Training Center Sample Number of Passes Sample Number of Passes Sample Number of Passes Location 0 10 25 50 Location 0 10 25 50 Location 0 10 25 50 CO X X X X CO X[b] X X X CO X X X X CI X X X X CI X[b] - - - CI X X X X SS X X X X SS X[b] - - X SS X X X X [a] Sample locations: CO = curve outside, CI = curve inside, SS = straight section. [b] Only one initial condition sample per figure-8 replication was taken at the Fort Riley site for each treatment.

Figure 3. Wind tunnel tray sample (122 cm long × 20 cm wide × 6 cm thick).

52 TRANSACTIONS OF THE ASABE (>100 μm) are deposited into a catch pan located under- entire length of the tunnel floor to create a boundary layer neath the sampler. This sampler was an integral part of the depth of approximately 0.3 m, based on previous research wind tunnel tray experiments and allowed dust emissions of with this wind tunnel (Mirzamostafa et al., 1998; Kohake et both suspension (<100 μm) and saltation plus creep al., 2010). The gravel was used to simulate roughness con- (≥100 μm) size ranges to be captured and measured. ditions more similar to the soil surface (1.6 mm random Two high-volume sampling blower motors were used roughness, as defined by Allmaras et al., 1966) than the during laboratory wind tunnel testing of the soil samples: smooth plywood tunnel floor. The roughness provided one model GBM2000H motor assembly for the mass flow boundary layer conditions within the air stream that better control (MFC) unit and one model GBM2000V motor as- replicate those found in actual field conditions (Kohake et sembly for the volume flow control (VFC) unit (General al., 2010). Note that the initial surface roughness varied Metal Works, Cleves, Ohio). The MFC unit had an adjusta- among the trays, and the roughness was further modified as ble flow range of 0.00944 to 0.02831 m3 s-1, with a flow the trays became armored (i.e., loose fines were removed) setpoint of 0.01888 m3 s-1 at 115 V and flow control accu- during testing. The aerodynamic roughness, excluding any racy of ±2.5% over a 24 h sampling period. The VFC unit standing residue effects, ranged from about 0.2 to 1.0 mm. was connected to a variable voltage controller, and the flow At the beginning of each test, the soil tray surface heights was adjusted to match the internal pitot tube (inside the slot were adjusted and leveled with the surrounding floor sur- sampler) pressures with the external pitot tube (outside the face using two floor jacks underneath the trays. The edge of slot sampler) pressures. the trays was either level with or below the surrounding soil These blower motors were used to create negative pres- and pea gravel so as to not impede or otherwise alter the sure within the slot-sampler system. At the beginning of the wind flow. experiments, one blower motor was voltage-adjusted to Testing was done for a period of 5 min at a mean wind -1 ensure that the total flow within the sampling system was speed of 14 m s , in accordance with the procedure of Ko- equivalent to the moving air stream within the wind tunnel hake et al. (2010), and was well above the observed thresh- during the tests. The setpoint resulted in a flow rate of old. The wind speed chosen was also near the upper limit of 0.0529 m3 s-1. The pressure was monitored in real time us- the tunnel. Wind speed was established at the centerline at ing pressure transducers connected to a computer. a height of 0.75 m above the tunnel floor using a pitot static Proper establishment of the upwind boundary layer con- tube coupled with a pressure transducer. ditions is an important factor to consider when using a wind Dust concentrations were measured with a Grimm spec- tunnel to estimate particulate emissions as a result of wind trometer (model 1.108, Grimm Aerosol Technik GmbH, erosion. Figure 5 shows a schematic of the wind tunnel. Ainring, Germany) from within the center of the ducting of The tunnel is a 12.2 m length × 1.2 m width × 1.5 m height, the isokinetic sampling train shown in figure 5. The Grimm push-type laboratory wind tunnel located at the USDA- spectrometer uses light scattering for single-particle counts, ARS Wind Erosion Laboratory in Manhattan, Kansas. In with a semiconductor laser as the light source. The scat- the upwind portion of the tunnel, directly following the fan, tered signal from a particle passing through the laser beam is a screen followed by a honeycomb structure. This setup is collected at approximately 90° by a mirror and trans- has been shown to decrease both lateral and longitudinal ferred to a recipient diode. The signal from the diode pass- turbulence (Rae and Pope, 1984). The boundary layer tures, after a corresponding reinforcement, through a multi- bulence was mechanical turbulence generated by the sur- channel size classifier. A pulse height analyzer then classi- face roughness. Spires extending upward from the floor fies the signal transmitted in each channel. The ambient air, to be analyzed, is drawn into the unit by an internal vol- surface are directly downwind of the honeycomb structure -1 to generate turbulence near the floor, which serves to ume-controlled pump at a rate of 1.2 L min , and results slightly increase the initial boundary layer. Pea-sized gravel are recorded every 6 s. The Grimm spectrometer provided was sieved to a size range of 5 to 7 mm and applied to the data for determination of 15 individual size fractions (>0.3, >0.4, >0.5, >0.65, >0.8, >1.0, >1.6, >2.0, >3.0, >4.0, >5.0,

Tray Honeycomb Slot sampler Spires

Figure 5. Schematic of the wind tunnel (12.2 m length × 1.2 m width × 1.5 m height).

58(1): 49-60 53 >7.5, >10.0, >15.0, and >20.0 μm) of the suspended dust in in soil texture. mg L-1. This allowed for the determination of the concen- For tracked vehicles (M1A1), the experimental design trations of particles in certain size ranges of interest, i.e., and treatment structure was exactly the same as with the <2.5 μm, <10 μm, and total dust (<20 μm). From the HMMWV, except that loam was excluded from soil texture measured concentrations, the volumetric flow rate through and the number of passes was 0, 1, 5, and 10. the Grimm spectrometer, and the duration of testing Heavy wheeled vehicles (M925A1) were tested only at (5 min), the total mass of dust (<20 μm) emissions (mg) one soil texture (loam). Therefore, the experimental design was determined. This value was divided by 0.0061 m2 was a randomized complete block design in which the (length of the tray times the width of the slot opening) to blocking factor was the replication and the treatment factor determine the dust emission potential (mg m-2) of the tray was number of passes (0, 2, 10, and 20). surface (eq. 1): All analyses were conducted using SAS software (ver. 9.3, SAS Institute, Inc., Cary, N.C.). Initial analyses assumed 1 n normality of responses and were conducted using the ECQt=×() ××Δ (1) A  i MIXED procedure. However, since residuals for each of the i=1 responses appeared rightwardly skewed (i.e., non-normal), a where generalized linear mixed model (GLMM) with a gamma E = emission per area (mg m-2) distribution and a log link function in the GLIMMIX proce- A = effective area of the tray (0.0061 m2) dure of SAS were used. See Stroup (2013) for technical de- n = number of concentration measurements during the 5 tails on GLMMs and the SAS online documentation min testing period (http://support.sas.com/documentation/onlinedoc/stat/) for technical details on the MIXED and GLIMMIX proce- Ci = PM concentration measurements every 6 s (mg m -3) dures. The selection of a gamma distribution ensured posi- Q = flow rate (m3 min-1) tive dust emission confidence limits while accounting for the heterogeneous variance experienced among soil tex- Δt = time interval (min). tures. Due to unbalanced subsampling, the Satterthwaite

denominator degree of freedom method was used. For light Background particulate concentration was measured be- wheeled and tracked vehicle analyses, F-tests were calcu- fore each run using the Grimm spectrometer. Additionally, lated for the main effects of soil texture and number of the wind tunnel was thoroughly cleaned after each run us- × ing high-velocity wind and rapping in conjunction with passes and for the soil pass interaction. For heavy compressed air cleaning of the sampling train while the wheeled vehicle analysis, the F-test was calculated only for high-volume samplers were running to clean any entrained number of passes. For all analyses, means and standard particles from the system. A settling period was allowed errors were also calculated for all effects. In addition, pair- after the high-velocity cleanout before starting the next wise comparisons were performed for the main effect sampling event. While the background concentration was means using the Tukey adjustment for Type I error and are not subtracted from the overall mass emission results, each reported here. test was run only after background concentration reached a non-detectable level. RESULTS AND DISCUSSION DATA ANALYSIS SOIL PROPERTIES Dust emission data were analyzed independently for the Soil properties measured in the upper 5 cm are presented three vehicle types (HMMWV, M1A1, and M925A1). For in table 1. Clay content for the study sites ranged from 28 each of these analyses, there were three replications for to 280 g kg-1, while sand ranged from 79 to 877 g kg-1, re- each vehicle/soil texture combination at each site. In addi- sulting in textures ranging from loamy sand to silty clay tion, tray samples, which were collected on the out- loam. Soils with less than 150 g kg-1 clay tend to have side/inside curves (CO, CI) and straight sections (SS) with- poorer aggregation and thus higher natural erodibility in each figure-8 location after an increasing series of traf- (Chepil, 1953, 1955). Organic matter contents ranged from ficking passes, were considered subsamples and averaged 9 g kg-1 for the warm and humid sandy soil at Fort Benning to obtain mean emissions from the entire figure-8 plot to 41 g kg-1 for the tallgrass prairie soils at Fort Riley. (fig. 2 and table 3). Since higher organic matter contents were also found in For light wheeled vehicles (HMMWV), the experi- soils with higher clay contents, the clay content is expected mental design was a completely randomized design with a to dominate the aggregation processes in these soils. The split plot. Soil texture (i.e., silty clay loam, silt loam, loam, soil samples were allowed to completely air-dry before the and loamy sand) was the fixed whole-plot treatment factor, wind tunnel tests; therefore, the samples had similar mois- and replication within soil texture was the random whole- ture contents, and soil moisture at the time of the tests was plot error term. Split-plot fixed effects included the main assumed to have no effect on results. However, the initial effect of number of vehicle passes (0, 10, 25, and 50) and moisture contents (i.e., prior to trafficking experiments) the interaction between soil texture and number of passes. were considerably lower for the loam (0.01 kg kg-1) and The split-plot error term was the random effect due to the loamy sand (0.02 kg kg-1) than for the silt loam (0.16 to interaction between number of passes and replication with- 0.17 kg kg-1) and silty clay loam (0.18 to 0.21 kg kg-1). Ini-

54 TRANSACTIONS OF THE ASABE tial bulk density ranged from 1.22 to 1.47 g cm-3. These es in vegetative cover due to trafficking on the soils at the values were within expected ranges and, because bulk den- sites where each vehicle was used. Note that the tracked sity has little effect on loose emissions of soil, they were vehicle was more destructive than either of the wheeled also assumed to have a minimal effect on results. vehicles with respect to vegetation cover. The denser prai- The higher water contents during the trafficking experi- rie grass vegetation and soils with higher clay content at ments and the ability of the Fort Riley soils to consolidate Fort Riley appeared to resist the destructive impact and better due to their higher clay contents may have limited shear forces in the curved regions of the multiple traffick- the emissions potential from these soils. If these soils were ing passes better than the Fort Benning and Yakima Train- present and tested under drier conditions with similar vege- ing Center vegetation and soils. Vegetation cover appears tation cover, like the other two sites, these soils may have to influence the emission of the soil surface, as expected generated emission levels more similar to the other soils. (figs. 6, 7, and 8). Additional experimentation is required to determine the actual effects, if any, that different soil water contents may LIGHT WHEELED VEHICLE have on emission potential due to repeated trafficking. Data were collected for the light wheeled vehicle (HMMWV) on four soil textures across the three experi- VEGETATION COVER mental sites. The difference in mean emission potential The trafficking passes degraded the vegetation cover for (<20 μm) for each testing parameter (i.e., pass and soil) all vehicles on all soils. Figures 6, 7, and 8 show the chang-

100 90 80 70 silty clay loam 60 silt loam 50 loamy sand 40 30 20 Field Vegetation Cover (%) Field Vegetation 10 0 01510 Vehicle Passes

Figure 6. Vegetative cover after vehicular traffic (M1A1 tank) at Fort Riley and Fort Benning sites.

100 90 80

70 silty clay loam 60 silt loam 50 loamy sand 40 30 loam

Field Vegetation Cover (%) Field Vegetation 20 10 0 0 102550 Vehicle Passes

Figure 7. Vegetative cover after vehicular traffic (HMMWV) on all sites.

58(1): 49-60 55 100 90 80 70 60 50 40 30 20 Field Vegetation Cover (%) Field Vegetation 10 0 0 2 10 20 Vehicle Passes

Figure 8. Vegetative cover after vehicular traffic (water tanker fire truck) on loam soil at Yakima Training Center. was significant (p < 0.05) for all emissions tests. Table 4 (silt loam), than the other two sites, which had 51.3% shows the pairwise comparisons for both soil texture and (loam) and 87.7% (loamy sand). Initial moisture content number of passes for all tests. All comparison tests showed (i.e., prior to trafficking experiments) also could have in- significant differences in emissions at the 5% level, except fluenced the behavior of the soil during trafficking and the the two soil textures at Fort Riley (silt loam and silty clay resulting emission potential. Note that the two soils at the loam). This result is not surprising, because the two soil Fort Riley site had initial moisture contents of 0.16 to textures have similar properties and were close in geo- 0.21 kg kg-1, whereas the other two sites had initial mois- graphical proximity. In addition, there was no significant ture contents of 0.01 to 0.02 kg kg-1. Although the effect of increase in emissions when the number of passes was dou- soil texture (combined with initial moisture content) was bled from 25 to 50 (p = 0.66). This result indicates that all more distinct than the vehicle passes, from the data pre- emission increases generally occurred between undisturbed sented in table 4, there is evidence that trafficking intensity conditions and 25 passes. Table 4 also shows that, from can influence increased emissions nearly as much as soil undisturbed conditions to 10 passes, emissions increased by texture alone. 357% on average for all light wheeled vehicle tests. From Figure 9 shows the total Grimm measured dust emission 10 to 25 passes, an additional 72% increase was observed. potential (particles <20 μm) for treatment interaction ef- Total dust emission potential as measured by the Grimm fects of soil texture and number of passes. The silty clay spectrometer was shown to be significantly influenced by loam and silt loam soil textures (with high initial moisture both trafficking passes and soil texture. In general, the soil contents) at the Fort Riley site had relatively low emission textures with more sand content were more susceptible to potential compared with the other sites. In addition, alt- emissions, as shown in table 4. As shown in table 1, the hough no significant (p > 0.05) increase in emissions was two soils at the Fort Riley site had significantly less sand detected between 25 and 50 passes, the loamy sand soil content at the surface, 7.9% (silty clay loam) and 10.2% continued to increase between these two treatment levels. Statistical analysis indicated significant (p < 0.05) effects Table 4. Pairwise comparisons of total dust emissions (<20 μm) from of both trafficking passes and soil texture, as well as their wind tunnel tests (Grimm spectrometer), light wheeled vehicle interaction. (HMMWV). Fort Benning (loamy sand) showed a 1,953% increase in Difference between Adjusted μ Soil 1 Soil 2 Estimate Treatments (%) p-Value emissions of total dust (<20 m) between undisturbed con- SiCL[a] SiL -0.60 82 0.13 ditions and 25 passes, and Yakima Training Center (loam) SiCL L -1.94 598 0.0002 had a 3,119% increase for the same treatment. Figure 10 SiCL LS -2.71 1396 <0.0001 illustrates the trend for the loam and loamy sand soils for SiL L -1.35 284 0.0021 μ μ μ SiL LS -2.11 722 <0.0001 total suspended particles (<20 m, <10 m, and <2.5 m). L LS -0.76 114 0.049 Pass 1 Pass 2 TRACKED VEHICLE 0 10 -1.52 357 <0.0001 For all one-way tests, the main effects of soil and traf- 0 25 -2.06 686 <0.0001 0 50 -2.27 868 <0.0001 ficking passes were significant (p < 0.05). The interaction 10 25 -0.54 72 0.031 effect between soil and pass was not significant (p > 0.05) 10 50 -0.75 112 0.002 for all tests, with the exception of emission potential of dust 25 50 -0.21 23 0.66 <2.5 μm. Note that the tracked vehicle testing represents [a] SiCL = silty clay loam (Fort Riley), SiL = silt loam (Fort Riley), L = loam (Yakima), and LS = loamy sand (Fort Benning). fewer data points than the light wheeled vehicle testing.

56 TRANSACTIONS OF THE ASABE 100000 ) -2

10000 m (mg m μ silty clay loam 1000 silty loam loam 100 loamy sand

Total Dust Emissions <20 Total 10

1 0102550 Vehicle Passes

Figure 9. Total dust emissions (<20 μm) from wind tunnel tests (Grimm spectrometer), light wheeled vehicle (HMMWV). Error bars represent 95% confidence limits of three replications.

4000

3500

) 3000 -2

2500 L - <2.5 µm L - <10 µm 2000 L - <20 µm 1500 LS - <2.5 µm

Dust Emissions (mg m 1000 LS - <10 µm LS - <20 µm 500

0 0 102550 Vehicle Passes

Figure 10. Trend similarities between particle sizes for the loam (L) and loamy sand (LS) for light wheeled vehicle (HMMWV).

This is due to the tracked vehicle being unavailable at the (<20 μm) potential was not significant (p > 0.05). One pos- Yakima site as originally planned. It is recommended that sible explanation for these results is that tracked vehicle further investigation into the effects of a tracked vehicle on Table 5. Pairwise comparisons of total dust emissions (<20 μm) from emission potential be conducted. wind tunnel tests (Grimm spectrometer), tracked vehicle (M1A1). Table 5 shows the pairwise comparison of the effects of Difference between Adjusted soil and trafficking passes on total dust emission (<20 μm) Soil 1 Soil 2 Estimate Treatments (%) p-Value potential. As seen with the light wheeled vehicle testing, SiCL[a] SiL -0.15 16 0.90 the two Fort Riley soils showed no significant (p > 0.05) SiCL LS -1.91 575 0.0028 SiL LS -1.76 481 0.0042 difference across all trafficking tests. Although there was a Pass 1 Pass 2 significant (p < 0.05) increase in emissions from undis- 0 1 -1.90 569 <.0001 turbed condition to any number of passes, no significant 0 5 -1.71 453 <.0001 (p > 0.05) difference was observed between 1, 5, and 0 10 -2.30 896 <.0001 1 5 0.19 -17 0.86 10 passes. In addition, unlike the results from the light 1 10 -0.40 49 0.38 wheeled vehicle testing, interaction effects between soil 5 10 -0.59 80 0.11 texture and number of vehicle passes on total dust emission [a] SiCL = silty clay loam (Fort Riley), SiL = silt loam (Fort Riley), and LS = loamy sand (Fort Benning).

58(1): 49-60 57 testing was completed on only three soil textures, and two (p > 0.05) for the tracked vehicle testing, the soil interac- of the three soils (silt loam and silty clay loam) were high tion effect for dust <20 μm was significant (p < 0.05). in clay content and relatively low in sand content (table 1), making them much less susceptible to emissions. In addi- HEAVY WHEELED VEHICLE tion, the tracked vehicle sheared the surface on the curves, Significant (p < 0.05) treatment effects were observed removing the top soil and most of the vegetation on the first for all heavy wheeled vehicle tests (fully loaded M925A1 pass (fig. 6). Subsequent passes resulted in similar behavior fire water tanker) at Yakima. After 20 passes of the heavy while exposing more moist soil below the surface, especial- wheeled vehicle, the soil surface showed a 5,276% increase ly at the Fort Riley site. In contrast, many more passes with in dust (<20 μm) emission potential over undisturbed con- the light wheeled vehicle were required to completely shear ditions. In contrast to the light wheeled and tracked vehi- through the vegetation and physically remove some of the cles, the emission increase between each pass level was top soil from the wheel tracks (fig. 7). significant (p < 0.05). Figure 11 illustrates the soil-pass interaction plot of total Figure 12 shows an increase in dust emission potential dust emissions (<20 μm) for the tracked vehicle. Although with each subsequent increase in vehicle passes for each most of the pass interaction effects were not significant treatment level up to 20 passes. The light wheeled vehicle

100000 ) -2

10000 m (mg m μ

1000 silty clay loam silty loam

100 loamy sand

Total Dust Emissions <20 Total 10

1 01510 Vehicle Passes

Figure 11. Total dust emissions (<20 μm) from wind tunnel tests (Grimm spectrometer), tracked vehicle (M1A1). Error bars represent 95% confidence limits of three replications.

7000 )

-2 6000

5000 m (mg m μ 4000

3000

2000

Total Dust Emissions <20 Dust Emissions Total 1000

0 0 2 10 20 Vehicle Passes

Figure 12. Total dust emissions (<20 μm) from wind tunnel tests (Grimm spectrometer), heavy wheeled vehicle (M925A1). Error bars represent 95% confidence limits of three replications.

58 TRANSACTIONS OF THE ASABE showed less increase in emissions at a higher number of multiple passes (>1) for the tracked vehicle, intensive vehi- passes; however, the heavy wheeled vehicle did not show cle training should be conducted, when possible, on lands the same increasing trend for the same soil texture. Note, that have been previously disturbed to reduce the amount of however, that the number of passes went up to only 20, land requiring remediation after trafficking. whereas the light wheeled vehicle used 50 passes as the upper level. This indicates that additional passes are neces- ACKNOWLEDGEMENTS sary to determine the maximum effect on the soil with the The authors acknowledge the Strategic Environmental heavy wheeled vehicle. Research and Development Program as well as the land and training coordinators at Fort Riley, Fort Benning, and Ya- kima Training Center. The authors also acknowledge the CONCLUSIONS contribution of Fred Fox in taking field tray samples and Multi-pass military vehicle trafficking experiments were Matt Kucharski and Youjie Xu in assisting with the wind conducted at three military training installations. Dust tunnel tests. emission potential was measured in a laboratory wind tunnel (with a Grimm spectrometer). The following conclusions were drawn from this research. REFERENCES Wind tunnel testing showed a significant impact of off- Allmaras, R. R., Burwell, R. E., Larson, W. E., & Holt, R. F. road trafficking passes on total dust emissions (<20 μm) for (1966). Total porosity random roughness of the interrow zone as most tests, except for the heavy wheeled vehicle. For both influenced by tillage. Conservation Report No. 7. Washington, D.C.: USDA Agricultural Research Service. the light wheeled and tracked vehicles, there seemed to be a Althoff, P. S., & Thien, S. J. (2005). Impact of M1A1 main battle threshold of maximum increase in emission potential. For tank disturbance on soil quality, invertebrates, and vegetation the light wheeled vehicle, from undisturbed conditions to characteristics. J. Terramech., 42(3-4), 159-176. 10 and 25 vehicle passes, emissions increased significantly http://dx.doi.org/10.1016/j.jterra.2004.10.014. (p < 0.05); however, when the number of passes doubled Althoff, P. S., Thien, S. J., & Todd, T. C. (2010). Primary and from 25 to 50, the increase was not significant (p > 0.05). residual effects of Abrams tank traffic on prairie soil properties. For the tracked vehicle, emissions increased significantly SSSA J., 74(6), 2151. http://dx.doi.org/10.2136/sssaj2009.0091. (p < 0.05) from undisturbed conditions to one pass, but Ashbaugh, L. L., Carvacho, O. F., Brown, M. S., Chow, J. C., when the number of passes was increased, the emissions Watson, J. G., & Magliano, K. C. (2003). Soil sample collection and analysis for the fugitive dust characterization study. Atmos. increase was not significant (p < 0.05). The heavy wheeled Environ., 37(9-10), 1163-1173. http://dx.doi.org/10.1016/S1352- vehicle did not exhibit exactly this behavior with the num- 2310(02)01022-1. ber of passes tested (2, 10, and 20), although that may be Chepil, W. S. (1953). Factors that influence clod structure and due to an insufficient number of passes being conducted erodibility of soil by wind: I. Soil texture. Soil Sci., 75(6), with this vehicle. 473.483. http://dx.doi.org/10.1097/00010694-195306000-00008. For the tracked vehicle, in general, emission potential Chepil, W. S. (1955). Factors that influence clod structure and increased in soils with higher sand content (and low initial erodibility of soil by wind: IV. Sand, silt, and clay. Soil Sci., moisture content). In addition, dust emissions increased 80(2), 155-162. http://dx.doi.org/10.1097/00010694-195508000- significantly from undisturbed conditions to any number of 00009. Diaz-Nigenda, E., Tatarko, J., Jazcilevich, A. D., Garcia, A. R., passes with the tracked vehicle. Caetano, E., & Ruiz-Suarez, L. G. (2010). A modeling study of The tracked vehicle was more destructive than either of Aeolian erosion enhanced by surface wind confluences over the wheeled vehicles with respect to vegetation cover, es- Mexico City. Aeolian Res., 2(2-3), 143-157 pecially when turning. This is due to the significant shear- http://dx.doi.org/10.1016/j.aeolia.2010.04.004. ing action in the curved regions and was likely intensified DoD. (2013). Base structure report: Fiscal year 2013 baseline. A by the extra weight of the tracked vehicle. The denser prai- summary of the Department of Defense’s real property rie grass vegetation and soils with higher clay content at inventory. Washington, D.C.: U.S. Department of Defense. Fort Riley appeared to resist the destructive impact and Retrieved from www.acq.osd.mil/ie/download/bsr/Base shear forces in the curved regions better than the Fort Ben- %20Structure%20Report%202013_Baseline%2030%20Sept%2 02012%20Submission.pdf. ning and Yakima Training Center vegetation and soils. The EPA. (2010). Fort Riley, Kansas: Site description. EPA No. heavy wheeled vehicle was more destructive of the vegeta- KS6214020756. Washington, D.C.: U.S. Environmental tion than the light wheeled vehicle at the Yakima Training Protection Agency. Retrieved from Center site, which can be attributed to the extra weight of www.epa.gov/region07/cleanup/npl_files/ks6214020756.pdf. the heavy wheeled vehicle. Whether the difference would Gee, G. W., & Dani, O. (2002). Particle-size analysis. In J. H. Dane be similar or different on other soils is unknown because & G. C. Topp (Eds.), Methods of Soil Analysis: Part 4. Physical the heavy wheeled vehicle was not used at the other sites Methods (pp. 255-293). Madison, Wisc.: SSSA. during this study. Gillies, J. A., Etyemezian, V., Kuhns, H., Nikolic, D., & Gillette, D. In general, soil texture (combined with initial moisture A. (2005). Effect of vehicle characteristics on unpaved road dust emissions. Atmos. Environ., 39(13), 2341-2347. contents) played an important role in dust emissions, but http://dx.doi.org/10.1016/j.atmosenv.2004.05.064. vehicle passes can produce an effect nearly as pronounced. Guoliang, L., Xuan, J., & Park, S. (2003). A new method to Because of the lack of a significant increase in emissions calculate wind proﬁle parameters of the wind tunnel boundary between 25 and 50 passes for light wheeled vehicles and

58(1): 49-60 59 layer. J. Wind Eng. Ind. Aerodynamics, 91(9), 1155-1162. Skidmore, E. L., & van Donk, S. J. (2003). Chapter 9: Soil erosion http://dx.doi.org/10.1016/S0167-6105(03)00057-6. and conservation. In D. W. Benbi, & R. Neider (Eds.), Kohake, D. J., Hagen, L. J., & Skidmore, E. L. (2010). Wind Handbook of Processes in the Soil-Plant Systems: Modeling erodibility of organic soils. SSSA J., 74(1), 250-257. Concepts and Applications (pp. 227-260). Binghamton, N.Y.: http://dx.doi.org/10.2136/sssaj2009.0163. Haworth Press. Marticorena, B., & Bergametti, G. (1997). Factors controlling Sloneker, L. L., & Moldenhauer, W. C. (1977). Measuring the threshold friction velocity in semiarid and arid areas of the amount of crop residue remaining after tillage. J. Soil Water United States. J. Geophys. Res., 102(D19), 23277-23287. Cons., 32(5), 231-236. http://dx.doi.org/10.1029/97JD01303. Stroup, W. (2013). Generalized Liner Mixed Models: Modern Mirzamostafa, N., Stone, L., Hagen, L. J., & Skidmore, E. L. Concepts, Methods, and Applications. Boca Raton, Fla.: CRC (1998). Soil aggregate and texture effects on suspension Press. components from wind erosion. SSSA J., 62(5), 1351-1361. Sweeney, M., Etyemezian, V., Macpherso, T., Nickling, W., Gillie, http://dx.doi.org/10.2136/sssaj1998.03615995006200050030x. J., Nikolich, G., & McDonald, E. (2008). Comparison of PI- NCDC. (2011). U.S. local climatological data. Asheville, N.C.: SWERL with dust emission measurements from a straight-line National Climatic Data Center. Retrieved from field wind tunnel. J. Geophys. Res., 113(F1), 1-12. www.ncdc.noaa.gov/most-popular-data#loc-clim. http://dx.doi.org/10.1029/2007JF000830. Rae, W. H., & Pope, A. (1984). Low-Speed Wind Tunnel Testing. Swift, R. S. (1996). Organic matter characterization. In J. M. Bartels New York, N. Y.: John Wiley and Sons. (Ed.), Methods of Soil Analysis: Part 3. Chemical Methods (3rd Retta, A., Wagner, L. E., Tatarko, J., & Todd, T. C. (2013). ed., pp. 1011-1069). Madison, Wisc.: SSSA. Evaluation of bulk density and vegetation as affected by military Van Pelt, R. S., Zobeck, T. M., Baddock, M. C., & Cox, J. J. vehicle traffic at Fort Riley, Kansas. Trans. ASABE, 56(2), 653- (2010). Design, construction, and calibration of a portable 665. http://dx.doi.org/10.13031/2013.42687. boundary layer wind tunnel for field use. Trans. ASABE, 53(5), 1413-1422. http://dx.doi.org/10.13031/2013.34911.

60 TRANSACTIONS OF THE ASABE